The current paradigm of using LLMs (and increasingly sophisticated monolithic systems that move beyond language in the sense of written human language, including code and images) rely on black boxes and Magic. How are CoEms different and how might they be used?
